CLI options - filter, no-delete mode, ignore-error mode

[aprudnev]

It can be very useful to have these things in both CLI options and yaml config

• filters, They exists in yaml but not in cli which is strange. It is much safer and easier to use cli for initial sync or to sync static data which do not requires 24x7 sync. cli have max file size but do not have filter.
• I had a few cases when directory contained 1 - 2 files without correct permissions. oikb failed instead of skipping such files.
• There are a cases when it is important to restrict sync from removing existing files. it can be critical KB or can be partial subdirectory update (when we do not want to update other directories).

System worked great, I synced 30K documents into 20+ KB collections, using Mistral OCR and pg_vector db. But system timed out few times (likely due to some memory leak in OWUI not oikb) and it tried to index jpg or png files which are not accepted by OWUI.

As we do not need 24x7 immediate sync thru webhooks, it is way easier and safer to run sync by cli instead of using the daemon. But, I do not see yet how can I use filters together with CLI. And as we have 20+ KB synced from 3 network file systems (cifs actually) and network may experience issues or some files can luck permission, better cli control (or a way to run cli but use additional config file) can be very useful.

[garciaden]

This would really help out with the Obisidian/RAG setup I've been using.

The oikb watch CLI option is what I actually want to use for my local Obsidian vault mount because notes are constantly changing, but the vault contains a lot of files that shouldn't be indexed (.obsidian/*, plugin data, cache files, images, etc.).

The YAML configuration supports include/exclude filters, but the watch command doesn't appear to expose the same filtering capabilities. It would be really helpful if watch could support include/exclude filters directly or maybe even use the same filter configuration as the YAML setup.

For example, I may only want to continuously sync .md and .pdf while excluding .obsidian/*, images, and other vault metadata/attachments.

Having consistent filtering behavior between one time syncs and watch mode would make local RAG setups much easier to manage and would prevent unnecessary indexing of files that aren't intended for the knowledge base.

[lm-gh]

Would be great if it would be possible to use watch also in the .oikb.yaml config.